Exploring multimodal implicit behavior learning for vehicle navigation in simulated cities
Fuente:
arXiv
Salvato in:
| Autori principali: | , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866909797278810112 |
|---|---|
| author | Antonelo, Eric Aislan Couto, Gustavo Claudio Karl Möller, Christian |
| author_facet | Antonelo, Eric Aislan Couto, Gustavo Claudio Karl Möller, Christian |
| contents | Standard Behavior Cloning (BC) fails to learn multimodal driving decisions, where multiple valid actions exist for the same scenario. We explore Implicit Behavioral Cloning (IBC) with Energy-Based Models (EBMs) to better capture this multimodality. We propose Data-Augmented IBC (DA-IBC), which improves learning by perturbing expert actions to form the counterexamples of IBC training and using better initialization for derivative-free inference. Experiments in the CARLA simulator with Bird's-Eye View inputs demonstrate that DA-IBC outperforms standard IBC in urban driving tasks designed to evaluate multimodal behavior learning in a test environment. The learned energy landscapes are able to represent multimodal action distributions, which BC fails to achieve. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_15400 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Exploring multimodal implicit behavior learning for vehicle navigation in simulated cities Antonelo, Eric Aislan Couto, Gustavo Claudio Karl Möller, Christian Machine Learning Artificial Intelligence Robotics Standard Behavior Cloning (BC) fails to learn multimodal driving decisions, where multiple valid actions exist for the same scenario. We explore Implicit Behavioral Cloning (IBC) with Energy-Based Models (EBMs) to better capture this multimodality. We propose Data-Augmented IBC (DA-IBC), which improves learning by perturbing expert actions to form the counterexamples of IBC training and using better initialization for derivative-free inference. Experiments in the CARLA simulator with Bird's-Eye View inputs demonstrate that DA-IBC outperforms standard IBC in urban driving tasks designed to evaluate multimodal behavior learning in a test environment. The learned energy landscapes are able to represent multimodal action distributions, which BC fails to achieve. |
| title | Exploring multimodal implicit behavior learning for vehicle navigation in simulated cities |
| topic | Machine Learning Artificial Intelligence Robotics |
| url | https://arxiv.org/abs/2509.15400 |